View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Am. J. Hum. Genet. 62:1052–1061, 1998

Genotype/Phenotype Correlation in Autosomal Recessive Lamellar Ichthyosis Hans Christian Hennies,1 Wolfgang Ku¨ster,2 Victor Wiebe,1 Alice Krebsova´,1,3 and Andre´ Reis1 1Institute of Human Genetics, Charite´, Humboldt University, Berlin; 2TOMESA Clinic for Allergy, Skin and Joint Diseases, and Rheumatism, Bad Salzschlirf, Germany; and 3Department of Medical Genetics II, University Hospital Motol, Charles University, Prague

Summary Introduction

Autosomal recessive lamellar ichthyosis is a severe con- Lamellar ichthyosis (LI) is a heterogeneous group of se- genital disorder of keratinization, characterized by var- vere hereditary disorders of keratinization. Marked dif- iable erythema of the whole body surface and by dif- ferences in the clinical picture and course of LI are ob- ferent scaling patterns. Recently, mutations have been served (Traupe 1989). In most cases, LI is inherited identified in patients with lamellar ichthyosis in the autosomal recessively; a rare form, however, of auto- TGM1 coding for keratinocyte , somal dominantly inherited LI has been described as well and a second locus has been mapped to 2. (Traupe et al. 1984). In the autosomal recessive type of We have now analyzed the genotype/phenotype corre- LI (McKusick 1994; MIM 242100), newborns often lation in a total of 14 families with lamellar ichthyosis. present a pergament-like collodion membrane. After los- Linkage analyses using microsatellites in the region of ing this encasement, patients show a characteristic gen- the TGM1 gene confirmed genetic heterogeneity. In pa- eralized scaling on the entire body, including scalp and tients not linked to the TGM1 gene, the second region flexural surfaces, and a variable erythema. Various clas- identified on chromosome 2 and a further candidate re- sifications of LI forms have been proposed, according gion on were excluded, confirming as to clinical, histological, and ultrastructural parameters well the existence of at least three loci for lamellar ich- (reviewed in Traupe 1989). Depending on the finding of thyosis. Sequence analyses of the TGM1 gene in families erythema, two major types of recessive LI were clinically compatible with linkage to this locus revealed seven dif- discriminated (Hazell and Marks 1985; Williams and ferent missense mutations, five of these unpublished so Elias 1986). The erythematous form of LI, also referred far, and one splice mutation. No genotype/phenotype to as nonbullous congenital ichthyosiform erythro- correlation for mutations in the TGM1 gene was found derma, is characterized by a strong redness of the entire in this group of patients, which included two unrelated body surface. Patients often also show an ectropion. In patients homozygous for the same mutation. Similarly, most cases, this form of LI is associated with a fine, no clear difference in the clinical picture was seen be- whitish scaling of the skin and marked palmoplantar tween patients with TGM1 mutations and those un- hyperkeratosis and hyperlinearity (Traupe 1989). In con- linked to the TGM1 locus. Comparison of genetic and trast, a nonerythematous form was described that often clinical classifications for patients with lamellar ichthy- exhibits large dark scales and a rough, barklike pattern osis shows no consistency and thus indicates that clinical of the skin (Williams and Elias 1986). A divergent clas- criteria currently in use cannot discriminate between the molecularly different forms of the disease. sification of patients with LI is obtained on the basis of histological and ultrastructural investigations. Five types of LI can be distinguished by assessment of ultrastruc- tural depositions in the stratum corneum and of the pat- terns of membranes and lamellar structures (Anton- Lamprecht 1992). The horny cells of the stratum corneum are charac- terized by a stable cell envelope formed of several com- Received August 5, 1997; accepted for publication February 27, 1998; electronically published April 17, 1998. ponents that are covalently bound to each other. These Address for correspondence and reprints: Dr. Andre´ Reis, Institut components include a number of proteins, such as in- fu¨ r Humangenetik, Charite´ Campus Virchow-Klinikum, Humboldt- volucrin, loricrin, and the small proline-rich proteins, Universita¨t zu Berlin, Augustenburger Platz 1, D-13353 Berlin, Ger- along with lipids on the outer surface of the cell enve- many. E-mail: [email protected] ᭧ 1998 by The American Society of Human Genetics. All rights reserved. lope. Keratinocyte transglutaminase (TGK), which cat- 0002-9297/98/6205-0009$02.00 alyzes the formation of covalent e-(g-glutamyl)lysine

1052 Hennies et al.: Genotype/Phenotype Correlation in LI 1053 bonds, is involved in the cross-linking of structural pro- (Compton et al. 1992) and with two microsatellites at teins in the upper granular layer. TGK is encoded by the anonymous loci, D14S64 and D14S264 (Dib et al. transglutaminase 1 gene (TGM1) on chromosome 1996), closely linked to TGM1. On chromosome 20, 14q11 (Kim et al. 1992; Phillips et al. 1992; Yamanishi microsatellite markers at D20S101, D20S107, and et al. 1992). Moreover, epidermal transglutaminase, D20S119 were analyzed. Genotyped markers on chro- which is encoded by the transglutaminase 3 gene mosome 2 are at D2S116, D2S325, D2S157, and (TGM3), is present in suprabasal keratinocytes. TGM3 D2S143 (Dib et al. 1996). Analyses were performed by has been localized to chromosome 20q11-q12 (Gentile use of an automated DNA sequencer (ALF, Pharmacia, et al. 1994; Wang et al. 1994). or 373, Perkin Elmer). Microsatellites were amplified in In patients with autosomal recessive LI, a lack of TGK 15-ml PCR mixtures containing 30 ng DNA, 0.33 mM expression has been shown (Hohl et al. 1993; Huber et of each primer, one of which was end-labeled with fluo- al. 1995a; Lavrijsen and Maruyama 1995), and linkage rescent dye, 0.1 mM of each dNTP, and 0.5 U Taq DNA of autosomal recessive LI to TGM1 has been described polymerase (Perkin Elmer) in 50 mM KCl, 10 mM Tris/

(Russell et al. 1994). Subsequently, mutations have been HCl (pH 8.0), and 1.5 mM MgCl2. Reactions consisted identified in the TGM1 gene in patients with LI (Huber of 27 cycles of 15 s at 94ЊC, 15 s at annealing temper- et al. 1995a; Parmentier et al. 1995; Russell et al. 1995). ature, and 30 s at 72ЊC. PCR products were separated In a number of families with autosomal recessive LI, on 6.6% polyacrylamide sequencing gels and were an- however, mutations in the TGM1 gene could be ruled alyzed by use of the Fragment Manager (Pharmacia) or out either by showing normal TGK activity (Huber et the Genotyper software (Perkin Elmer). al. 1995b; Lavrijsen and Maruyama 1995) or by ex- cluding linkage to TGM1 (Parmentier et al. 1995; Bale Linkage Analysis et al. 1996). Furthermore, a second locus for LI (ICR2B) Linkage analyses were performed with the computer has recently been mapped, in three consanguineous fam- programs in the LINKAGE package, version 5.1 (La- ilies from Morocco, to a 6.6-cM interval between the throp et al. 1984). Independent segregation of the phe- markers at D2S325 and D2S137 on chromosome 2q33- notype in question and a locus analyzed is considered q35 (Parmentier et al. 1996). significant for LOD scoresZ ! Ϫ2 . Autosomal recessive Here we present the investigation of a total of 14 inheritance with complete penetrance was assumed in families with clinically different forms of autosomal re- all LI families investigated. cessive LI. Our findings confirm that LI is genetically heterogeneous but do not show consistency of the mo- Mutation Analysis lecular results with clinical classifications. Mutations in the TGM1 gene were identified in a number of LI pa- The translated exons 2–15 of the TGM1 gene were tients, but no genotype/phenotype correlation could be amplified by PCR with intronic primers designed ac- derived. cording to published TGM1 sequences (Kim et al. 1992; Phillips et al. 1992). DNA was amplified by PCR as Material and Methods described above. For SSCP analyses, ∼10% of the vol- ume of the PCR sample was denatured and separated Patients with LI on 8% polyacrylamide gels according to the procedure described by Orita et al. (1989). Electrophoresis was A total of 14 families with autosomal recessive LI were performed at 40 W and 4ЊCor15ЊC for 1.5–6 h. PCR analyzed clinically and molecularly. Thirteen of these products were sequenced directly either by use of a cycle- originate from Germany, and one is of Moroccan ex- sequencing protocol (Amersham) or by solid-phase se- traction. In two families, only one patient each and the quencing with magnetic beads (Dynal) and T7 DNA patient’s parents were available. All others are two- or polymerase (Pharmacia). Sequencing products were an- three-generation families with one or two affected in- alyzed on an automated DNA sequencer (Pharmacia). dividuals. The family from Morocco is consanguineous, A detailed description of the transglutaminase activity an unresolved relationship is remembered in one of the assay will be given elsewhere (M. Huber and D. Hohl, German families, and the parents of all other patients personal communication). In brief, frozen skin sections are unrelated. were incubated in 100 mM Tris/Cl pH 8, 1% BSA for 30 min, to block nonspecific binding, and then in 100 Microsatellite Analysis mM Tris/Cl pH 8, 5 mM CaCl2, and 12 mM monodan- DNA was prepared according to standard methods, sylcadaverine for 1 h, to detect transglutaminase activity. from blood samples drawn from consenting individuals. For a negative control in normal human skin, EDTA Linkage analysis on chromosome 14 was performed was added to a final concentration of 20 mM. After the with the microsatellite located within the TGM1 gene transglutaminase reaction was stopped in PBS/10 mM 1054 Am. J. Hum. Genet. 62:1052–1061, 1998

EDTA, sections were incubated with rabbit antidansyl Patients PB and EB.—The patients are brothers, 10 antibody (1:100) (Aeschlimann et al. 1993) and mono- and 17 years of age. Family history was unremarkable, clonal anti-TGK antibody B.C1 (1:1) (Thacher and Rice the patients have two unaffected siblings, and the parents 1985) in 12 % BSA/PBS for 3 h. Sections were then were not consanguineous. Both brothers were born as incubated with biotinylated horse antimouse antibody collodion babies. Their clinical presentation was iden- (1:100) in 12% BSA/PBS and normal horse serum (1: tical, with absent or only slight erythema; platelike, 100) for 30 min and afterwards with fluorescein iso- large, dark brown scales of the whole skin, including thiocyanate–labeled swine antirabbit antibody (1:40) scalp, face, and the flexural folds; slight ectropion; mod- and Texas Red streptavidin (1:400) in 12% BSA/PBS for erate hyperkeratosis of palms and soles; and 30 min. hypohidrosis. Patients FS and JS.—The children are identical female Results twins, aged 6 years. Consanguinity of the parents is not discerned, but both grandparents originate from small neighboring villages. There is one healthy younger Clinical Investigations brother. Both children were born in the 31st wk of ges- Fourteen families with autosomal recessive LI, from tation, as collodion babies with ectropion, eclabium, and Germany and Morocco, were investigated clinically, mo- contractures of fingers and toes. Clinical investigation lecularly, and immunohistochemically (fig. 1 and table presented identical findings in both children, with 1). The clinical picture of the patients analyzed was marked erythema, fine white scaling of the entire skin, highly variable. Most were born as collodion babies, and including the scalp, a darker, grey-brownish scaling only many showed joint contractures and palmoplantar hy- on the back of the feet, and slight palmoplantar kera- perkeratosis. Nine unrelated patients suffered from er- tosis. Both children have a marked hypohidrosis. ythema, accompanied—in the more severely affected Patient JL.—Both parents and three of the grandpar- cases—by ectropion, whereas five patients from different ents were born in the same town, but consanguinity of families presented a nonerythrodermic form of LI. The the parents is not known. There are no siblings. At birth, type of scaling was highly variable, ranging from fine patient JL had no collodion membranes and no marked scales of fair color to dark brownish, polygonal, barklike erythema. The clinical findings in the 35-year-old man scales (table 1). The LI phenotype breeds true, in all showed a generalized erythema of moderate intensity families where ascertainable. Thirteen LI patients from and a fine scaling of the skin, lighter on the arms and nine different families of German origin, who were re- the trunk and darker on the lower legs. There is hy- vealed to carry inactivating mutations in the TGM1 pohidrosis and marked keratosis on palms and soles. gene, are described in more detail below. Similar disorders are not remembered in the family. Patient NE.—The parents of patient NE are not related Patient KM.—The parents are not related, and the pa- and have a healthy younger son. The affected girl was tient, a boy, is their only child. He was born in the 34th born at term, as a collodion baby with ectropion. There wk of gestation, with generalized erythema; probably was a remarkable improvement of the skin condition with collodion membrane; partially with platelike ker- after rupturing and solving of the collodion membranes atoses similar to those seen in harlequin ichthyosis; ec- some weeks after birth. At the age of 1 year, only a tropion; and eclabium. Clinical examination at the age minimal, superficial, white scaling of the thighs remained of 3 years revealed a marked erythroderma; moderate visible. There were no clinical signs of generalized ery- ectropion; light grey-brown scales of the entire skin with thema, scaling, ectropion, or palmoplantar kera- marked, platelike scaling of the scalp; moderate pal- toderma. Similar disorders are not known within the moplantar keratoderma; and hypohidrosis. Keratiniza- family. tion disorders are unknown in the family. Patient SW.—The female patient has one older Patient VB.—The 5-year-old girl has a healthy older brother, one older sister, and one daughter, all of whom brother. She was born as a collodion baby. Clinical find- are healthy. Ichthyosis was present in the patient at birth, ing showed a moderate erythema; brown scales, espe- with large brown scaling without erythema. During cially on the lower limbs; platelike light brown hyper- childhood, the color of scales was light brown to dark keratosis of the scalp; marked palmoplantar kera- brown, but later, scales were of black-brown, especially toderma; and generalized hypohidrosis. Similar disor- at flexural folds. Clinical examination of the patient at ders are not remembered in the family. Consanguinity the age of 33 years presented no visible erythema of the of the parents is not known, but both grandparents come skin, light scales on the entire surface of the body, no from a small village of about 3,000 inhabitants. ectropion, and normal skin of palms and soles. There Patients AW and PW.—The patients are siblings and were no ichthyotic disorders in the family history and have one healthy older brother. Their parents are healthy no consanguinity of the parents. and not related. Both children were born as collodion Hennies et al.: Genotype/Phenotype Correlation in LI 1055

Figure 1 Pedigrees of 14 families with autosomal recessive LI and haplotypes aligned from genotypes at TGM1, D14S264, and D14S64. The phases in the parents are unknown in most families and deduced in the children. Upper panel: Families analyzed for mutations in the TGM1 gene. Five families—FS, NE, SU, VB, and PB—are compatible with linkage to TGM1. Family AW is only partly informative for the markers analyzed but the base substitution 4294GrA in the TGM1 gene, which results in mutation G382R, reveals a recombination between D14S264 and TGM1 (arrowhead). Simplex families KM and JL are not suitable for linkage analyses. The splice mutation A3447G in the TGM1 gene has been identified seven times in families SU, VB, SW, JL, and PB. The mutation has been found on the same haplotype, 222–174–130, at the loci D14S264, TGM1, and D14S64 in six cases. This haplotype, coded as 5–3–3, is boxed in the figure. The corresponding haplotype in family PB, 222–174–128, is coded as 5–3–4. Lower panel: Families with LI not linked to TGM1. Linkage has been excluded in five families by the identification of obligatory recombination events. babies. Clinically, they presented an almost identical Patients SU and JU.—Both are siblings and the only finding, with moderate erythema of the entire skin; slight children of healthy parents. A distant relationship of the ectropion; palmoplantar keratoderma; and a platelike parents was reported, but consanguinity could not be brown scaling of the complete skin that is lighter on the explicitly determined. The patients noted that they had scalp, arms, and trunk but darker on the lower legs and ectropions, as children, and hypohidrosis until the age in the great body folds. Furthermore, there was a marked of 16 years. Clinical findings at the ages of 30 and 31 hypohidrosis. years showed identical skin lesions, with large scales of 1056 Am. J. Hum. Genet. 62:1052–1061, 1998

Table 1 Summary of Clinical Features, Linkage, and Mutation Data for 14 Families with LI

CLINICAL FEATURES LINKAGEa CONSAN- MUTATIONS IN TGK FAMILY ORIGIN GUINITY Erythema Scaling TGM1 ICR2B TGM2/3 TGK ACTIVITY NE* German ϪϪWhitish, fine ϩ NT NT R323Q NA SW* German ϪϪFair color NI NT NT A3447G/A3447G NA PB* German ϪϪDark, platelike ϩ NT NT R315C/A3447G Ϫ FS* German ϪϩWhitish, fine ϩ NT NT V518M NA JL* German ϪϩFine NA NA NA G144R/A3447G Ϫ Light brown, KM* German Ϫϩmedium-size NA NA NA V518M/V518M Ϫ VB* German ϪϩPlatelike ϩ NT NT A3447G/A3447G NA AW* German ϪϩPlatelike NI NT NT R225H/G382R NA SU* German ? ϩ Barklike ϩ NT NT R142H/A3447G NA

NA Moroccan ϩϪDark ϪϪ Ϫ NT NA HG German ϪϪPlatelike ϪϪ Ϫ NT NA AE German ϪϩFair color, fine ϪϪ Ϫ NT NA BK German ϪϩFair color, fine ϪϪ Ϫ NT NA Light brown, JK German Ϫϩmedium-size Ϫ NI Ϫ NT NA

NOTE.—A detailed clinical description of the patients with mutations in TGK, indicated by an asterisk (*), is given in the text. Linkage analyses were not possible in two simplex families, but mutation analyses in the TGM1 gene were performed in patients from these families as well. NI ϭ not informative; NA ϭ not available; and NT ϭ not tested. A plus sign (ϩ) indicates presence; a minus sign (Ϫ) indicates absence; and the question mark (?) indicates that a possible consanguinity could not be clarified. a Assessment for linkage to candidate regions with microsatellites on 14 (TGM1), 2 (ICR2B), and 20 (TGM2 and TGM3); see text for details. the scalp; fine, light scaling of the face; and large brown data for TGM1, D14S64, and D14S264 (fig. 1 and table scales of trunk and limbs, with a barklike pattern of the 1). However, these results give only a clue to linkage, skin. A patchlike erythema was seen, along with hyper- since single families taken into account are too small to keratosis of palms and soles and clubbing of the nails. give firm evidence. For the loci D14S264 and D14S64, neither recombination to TGM1 (Russell et al. 1994) Linkage Studies nor to one another (Dib et al. 1996) has been reported so far. In family AW, haplotypes could not be aligned at In 12 families with autosomal recessive LI—one of these three loci because the mother is uninformative for these consanguineous, with four to six family members the microsatellite at TGM1. A mutation identified in from two or three generations—linkage analyses were exon 7 of the TGM1 gene in the mother and her affected performed by use of microsatellites in the chromosomal children, however, revealed a recombination between region of TGM1 on chromosome 14q (Kim et al. 1992; Phillips et al. 1992; Yamanishi et al. 1992). The re- D14S264 and TGM1 in this family (fig. 1). No further maining two families were simplex families and not suit- recombination event was found in this interval, in the able for linkage analyses. To directly assess the segre- families described here. Family SW was not informative gation at the locus TGM1, we analyzed the highly for the three markers. Five families show obligatory re- informative microsatellite in intron 14 of the TGM1 combination events between the LI phenotype and the gene (Compton et al. 1992). In addition, we inviestigated three microsatellites close to the TGM1 locus (fig. 1). microsatellites at the anonymous loci D14S64 and The families shown not to be linked to TGM1 were D14S264. A cDNA identified by the Whitehead Insti- additionally analyzed for linkage to two other candidate tute, WI-9158, is identical to the TGM1 coding se- regions for LI. It has been demonstrated by FISH that quence. D14S64 has been localized on the distal side the locus TGM3 is located in close proximity to the and D14S264 on the proximal side of WI-9158 by anal- tissue transglutaminase gene locus (TGM2) on the long ysis of YAC contigs (Hudson et al. 1995, with supple- arm of chromosome 20 (Wang et al. 1994). The cDNA mentary data from the Whitehead Institute/MIT Center WI-7847 of the Whitehead Institute, which is identical for Genome Research, Mapping to the TGM2 coding sequence, has been mapped to the Project). 1.2-cM interval between D20S107 and D20S99 by ra- Five families were compatible with linkage of LI to diation hybrid mapping (Schuler et al. 1996). The mi- the locus TGM1 according to genotype and haplotype crosatellite markers at D20S101, D20S107, and Hennies et al.: Genotype/Phenotype Correlation in LI 1057

Figure 2 Pedigrees of 5 families with autosomal recessive LI not linked to TGM1 and haplotypes in candidate regions for LI on chromosomes 2 and 20. The phases in the parents are unknown, in most families, and are deduced in the children. Upper panel: Haplotypes aligned from genotypes at D2S116, D2S325, D2S157, and D2S143 in the region of the second locus for LI, ICR2B. This locus has recently been mapped between the markers at D2S325 and D2S137 (Parmentier et al. 1996). Linkage to ICR2B could be excluded in families BK, HG, and AE. In family NA, linkage can most likely be excluded by the lack of homozygosity by descent. Family JK is not informative because of a recombination between D2S157 and D2S143 (arrowhead). Lower panel: Haplotypes aligned from genotypes at D20S101, D20S107, and D20S119 in the region of the TGM2 and TGM3 . Linkage to this region could be excluded in four families. In family NA, again, linkage can most likely be excluded by the lack of homozygosity by descent.

D20S119 cover a 13-cM region containing the TGM2 and exclusion of linkage is confirmed by haplotype anal- and TGM3 gene loci. Although the number of probands yses in four of the families (fig. 2). Linkage can most in each of the five LI families analyzed is too small for likely be excluded by lack of homozygosity in the con- obtaining effective LOD scores, the transglutaminase sanguineous Moroccan family NA, in which both grand- gene loci on chromosome 20 could be excluded from fathers are brothers and both grandmothers are sisters. linkage. Obligatory recombination events were identified Furthermore, microsatellites at D2S116, D2S325, for at least two of the chromosome 20 markers analyzed, D2S157, and D2S143 were genotyped. These loci are 1058 Am. J. Hum. Genet. 62:1052–1061, 1998 located on chromosome 2q33-q35 in the region to which the second locus for LI, ICR2B, has recently been mapped (Parmentier et al. 1996). Linkage to this locus could also be excluded in at least three of the families not linked to TGM1 (fig. 2 and table 1). In family NA, again, only lack of homozygosity could be demon- strated, and in family JK, which shows a recombination between D2S157 and D2S143, linkage to the second LI locus cannot be excluded (fig. 2).

Mutation Studies Mutation analyses in the TGM1 gene were performed in nine families with autosomal recessive LI by SSCP Figure 3 Schematic depiction of the putative secondary structure analyses and direct sequencing of TGM1 exons. Seven of the TGK polypeptide, based on the secondary structure of blood coagulation factor XIII A (Takahashi et al. 1986; Yee et al. 1994) different missense mutations and one splice mutation whose primary structure is highly similar to TGK. The positions of were found in TGK (fig. 3 and table 1). Besides muta- mutations identified in patients with autosomal recessive LI described tions R142H (Russell et al. 1995) and R323Q (Huber here are marked by arrows. In the case of the splice mutation A3447G, et al. 1995a), five novel missense mutations not de- the position of the sequence insertion is indicated, which is caused by scribed so far were detected: G144R (nucleotide failure in excision of intron 5 and results in a premature stop codon. exchange 1494GrA in TGM1, according to Kim et al. 1992), R225H (2730GrA), R315C (3515CrT), recessive LI, we have analyzed 14 families with LI, from G382R (4294GrA), and V518M (7915GrA). These Germany and Morocco. Twelve of these were suitable mutations were not present on 100 chromosomes of un- for linkage studies, and five families could be excluded affected control individuals of German origin. The splice from linkage to TGM1 (table 1). Two families were un- mutation affecting the 3 splice site of intron 5, A3447G informative, but the remaining five families were com- (according to Kim et al. 1992), has been described by patible with TGM1 linkage. The families not linked to Huber et al. (1995a) as A3366G. This mutation was the TGM1 gene were additionally assessed for linkage identified on seven chromosomes. Two LI patients, SW to two further candidate regions. Linkage of LI to the and VB, are homozygous for this splice mutation, loci TGM2 and TGM3 on chromosome 20, however, whereas patients JL, PB/EB, and SU/JU are compound could be excluded. A second locus for autosomal reces- heterozygous for A3447G and one missense mutation. sive LI, ICR2B, has been localized to chromosome 2q Remarkably, six of seven chromosomes carrying muta- (Parmentier et al. 1996). This locus could be clearly ruled tion A3447G have the same haplotype, 222–174–130, out as well in three of the families. Consanguineous fam- aligned from genotypes at D14S264, TGM1, and ily NA, from Morocco, can most likely be excluded by D14S64 (fig. 1). In the seventh case, family PB, the cor- lack of homozygosity for the markers at ICR2B, which responding haplotype is 222–174–128 (i.e., it deviates has originally been identified in Moroccan families. at D14S64 by only one [CA] repeat). The respective It has been noted elsewhere that TGK expression is mutations were also found in the parents of affected in fact detectable in a number of LI patients (Huber et individuals, and, in all cases, both parents of homozy- al. 1995b; Lavrijsen and Maruyama 1995), and a total gous patients were shown to be heterozygous for the of 15 families not linked to the TGM1 region have been mutation. This finding indicates that all cases are truly described so far (Parmentier et al. 1995; Bale et al. 1996). familial and recessively inherited. To confirm that the The findings presented here substantiate the genetic het- TGM1 mutations detected in affected individuals really erogeneity of LI and demonstrate that mutations re- impair the function of the enzyme, we investigated skin sulting in a LI phenotype must exist in at least three sections immunohistochemically from three LI patients different genes. There is no obvious candidate for a new (PB, JL, and KM) from whom skin biopsies were avail- LI locus, and we have therefore embarked on a ge- able (table 1). All of these patients showed markedly nomewide search for further loci. decreased activity of TGK. Mutation Studies Discussion SSCP and sequencing analyses of the TGM1 gene were performed in LI patients, and a TGK assay was carried Linkage Studies out in three cases in which a skin biopsy was available. To address the question whether a genotype/pheno- Eight different mutations were found, including seven type correlation can be found in patients with autosomal missense mutations and one splice mutation (fig. 3 and Hennies et al.: Genotype/Phenotype Correlation in LI 1059 table 1). Among these, mutations R142H and R323Q or an older recombination between TGM1 and D14S64. have been described elsewhere (Huber et al. 1995a; Rus- Patients SU and JU are compound heterozygotes for sell et al. 1995). Mutation G144R resides in a region of TGM1 mutations, although their parents are related. No TGK that obviously exhibits a mutational hot spot. A details about this distant relationship are known, how- total of four different mutations have been identified in ever, and the mutations were found on different hap- the residues R142, R143, and G144 (Huber et al. 1995a; lotypes, one of these the common haplotype carrying the Russell et al. 1995). The lack of TGK expression found recurrent mutation A3447G. In two cases, the patient in patient JL strengthens the likelihood that G144R is NE and the sisters FS and JS, only one mutation in the an inactivating mutation. It resides in a segment sup- TGM1 gene was identified. The apparent absence of the posed to be crucial for the correct structure and function second TGM1 mutation is presumably due to the in- of the enzyme (Kim et al. 1994). Moreover, the residues sufficient sensitivity of the SSCP analysis. In both of these at positions 142–145 are extremely conserved between cases, the mutation found has also been described in various and among many species. The other LI patients with absent TGK activity. All the mu- mutation R225H changes a moderately conserved res- tations identified in the TGM1 gene can be assumed to idue, which is also part of the presumed b-sandwich result in impairment of TGK activity and are therefore domain of TGK. It might also act by altering the folding the cause for the phenotype of LI. of the protein. Missense mutation R315C again affects a highly conserved residue. Furthermore, a mutation has Genotype/Phenotype Correlation been found in the arginine residue at position 252 of another transglutaminase, blood coagulation factor XIII Autosomal recessive LI is characterized by remarkable A, causing inherited factor XIII deficiency (Mikkola et clinical heterogeneity. An erythematous and a nonery- al. 1996). The primary structure of factor XIII A (Tak- thematous type of LI were distinguished clinically (Ha- ahashi et al. 1986) is highly similar to the structure of zell and Marks 1985; Williams and Elias 1986). More- TGK, and the position 252 in factor XIII A corresponds over, patients might present different patterns and colors to residue 315 in TGK. of the scales, they might or might not have an involve- Mutation G382R significantly alters a conserved res- ment of palmoplantar hyperkeratosis, and most but not idue. Glycine or alanine residues are found at the cor- all were born as collodion babies. We have therefore responding positions in other transglutaminases, but no compared two genetically differentiated groups of LI pa- charged side chains, such as those in the arginine residue, tients: one group comprising five unrelated patients ex- are introduced here. Moreover, G382R affects a residue cluded from linkage to the TGM1 gene (referred to as close to C377 that is part of the supposed catalytic triad non–TGM1 LI patients), and one group of nine patients of TGK (Yee et al. 1994). It belongs to the highly con- with LI unambiguously caused by TGM1 gene muta- served region around the active site of the enzyme (Tak- tions (TGM1 LI). ahashi et al. 1986). Remarkably, mutation V518M re- The group of non–TGM1 LI patients included ery- sides much more distal in the TGM1 gene than any other thematous as well as nonerythematous cases, and they mutation described so far. It affects a residue even distal presented the whole range between fine scales of fair, to the active site and the potential catalytic triad (fig. 3), whitish color and large, dark brown scales (table 1). and this residue is only moderately conserved in trans- These differences in the clinical picture could still be glutaminases. The mutation, however, has been identi- explained by different genes causing the disorder. The fied on a total of three chromosomes, patients FS being entire spectrum of LI variants, however, was also seen heterozygous and KM being homozygous for this mu- in TGM1 LI patients (table 1). Patients with erythema tation. Moreover, the lack of TGK activity in patient from both groups often showed fine and light scaling KM confirms that this is also a truly inactivating (six of nine); patients with the nonerythematous form mutation. of LI were mostly affected by large, dark brown scales Splice mutation A3447G results in premature stop co- (three of five). don, as described by Huber et al. (1995a). It has been In seeking a genotype/phenotype correlation for spe- observed on the same haplotype in six cases, which cific mutations in the TGM1 gene, we were able to com- points at a common origin of the chromosomes with pare two patients who carry the same TGM1 genotype. mutation A3447G and suggests that this is a rather old Both the patients VB and SW are homozygous for the mutation. In the brothers PB and EB and in their mother, splice mutation A3447G, but they differed in their clin- A3447G was identified on a haplotype identical at ical picture. VB suffers from a moderate erythema. She D14S264 and TGM1. The allele at D14S64 is different shows brownish scales, particularly affecting the lower by only 2 bp (i.e., one [CA] repeat). This chromosome limbs, and marked palmoplantar hyperkeratosis. On the could therefore present either the same ancient haplo- contrary, SW presented no visible erythema, scaling of type, with a mutation in the microsatellite at D14S64, fair color on the entire surface of the skin, but no in- 1060 Am. J. Hum. Genet. 62:1052–1061, 1998 volvement of palms and soles. The size and color of Millasseau P, et al (1996) A comprehensive genetic map of scales in patient SW, however, seem to have changed over the human genome based on 5,264 microsatellites. Nature the course of time, even in the absence of treatment with 380:152–154 retinoids. These findings, together with the clinical var- Gentile V, Davies PJ, Baldini A (1994) The human tissue trans- iability seen within the entire group of TGM1 LI pa- glutaminase gene maps on chromosome 20q12 by in situ fluorescence hybridization. Genomics 20:295–297 tients, contradict a genotype/phenotype correlation for Hazell M, Marks R (1985) Clinical, histologic, and cell kinetic mutations in the TGM1 gene. Particularly, this is also discriminants between lamellar ichthyosis and nonbullous true for the phenotypical differentiation between the er- congenital ichthyosiform erythroderma. Arch Dermatol ythematous and the nonerythematous forms of LI. 121:489–493 In summary, no characteristic differences could be de- Hohl D, Huber M, Frenk E (1993) Analysis of the cornified tected in the clinical presentation between the TGM1 LI cell envelope in lamellar ichthyosis. Arch Dermatol 129: and non–TGM1 LI patients. Each group displayed the 618–624 whole spectrum of symptoms for LI, such as erythem- Huber M, Rettler I, Bernasconi K, Frenk E, Lavrijsen SP, Ponec atous and nonerythematous forms, various scaling pat- M, Bon A, et al (1995a) Mutations of keratinocyte trans- terns, and existence or lack of palmoplantar kerato- glutaminase in lamellar ichthyosis. Science 267:525–528 derma. Hence, there are currently no conclusive criteria Huber M, Rettler I, Bernasconi K, Wyss M, Hohl D (1995b) Lamellar ichthyosis is genetically heterogeneous—cases with for the clinical differentiation of LI types corresponding normal keratinocyte transglutaminase. J Invest Dermatol to their genetic classification. There are many gene prod- 105:653–654 ucts involved in the terminal differentiation of the epi- Hudson TJ, Stein LD, Gerety SS, Ma J, Castle AB, Silva J, dermis, and mutations in quite a few of these can be Slonim DK, et al (1995) An STS-based map of the human envisaged as a cause for the symptoms seen in LI. Only genome. Science 270:1945–1954 the identification of further genes for LI will allow a Kim IG, McBride OW, Wang M, Kim SY, Idler WW, Steinert refined genetic classification of autosomal recessive LI. PM (1992) Structure and organization of the human trans- Correspondingly, as those findings give further insight glutaminase 1 gene. J Biol Chem 267:7710–7717 into differentiation processes in the skin, they will pro- Kim SY, Kim IG, Chung SI, Steinert PM (1994) The structure vide the basis for further understanding of the etiology of the transglutaminase 1 enzyme: deletion cloning reveals of LI. domains that regulate its specific activity and substrate spec- ificity. J Biol Chem 269:27979–27986 Lathrop GM, Lalouel JM, Julier C, Ott J (1984) Strategies for Acknowledgments multilocus linkage analysis in humans. Proc Natl Acad Sci USA 81:3443–3446 We are indebted to all the patients and their relatives who Lavrijsen AP, Maruyama T (1995) Absent transglutaminase contributed to this study. We are most grateful to Marcel TGK expression in two of three patients with lamellar ich- Huber and Daniel Hohl, Lausanne, for performing the TGK thyosis. Arch Dermatol 131:363–364 activity assay, and we wish to thank Silke Appel, Berlin, who McKusick VA (1994) Mendelian inheritance in man, 11th ed. participated in SSCP analyses, and Martin Jung, Berlin, for his Johns Hopkins University Press, Baltimore support in automated linkage analysis. This work was sup- Mikkola H, Yee VC, Syrjala M, Seitz R, Egbring R, Petrini P, ported by the Deutsche Forschungsgemeinschaft. Ljung R, et al (1996) Four novel mutations in deficiency of coagulation factor XIII: consequences to expression and References structure of the A-subunit. Blood 87:141–151 Orita M, Suzuki TS, Hayashi K (1989) Rapid and sensitive Aeschlimann D, Wetterwald A, Fleisch H, Paulsson M (1993) detection of point mutations and DNA polymorphisms using Expression of tissue transglutaminase in skeletal tissues cor- the polymerase chain reaction. Genomics 5:874–879 relates with events of terminal differentiation of chondro- Parmentier L, Blanchet-Bardon C, Nguyen S, Prud’homme J- cytes. J Cell Biol 120:1461–1470 F, Dubertret L, Weissenbach J (1995) Autosomal recessive Anton-Lamprecht I (1992) The skin. In: Papadimitriou JM, lamellar ichthyosis: identification of a new mutation in Henderson DW, Spagnolo DV (eds) Diagnostic ultrastruc- transglutaminase 1 and evidence for genetic heterogeneity. ture of non-neoplastic diseases. Churchill Livingstone, Ed- Hum Molec Genet 4:1391–1395 inburgh, pp 459–550 Parmentier L, Lakhdar H, Blanchet-Bardon C, Marchand S, Bale SJ, Russell LJ, Lee ML, Compton JG, DiGiovanna JJ Dubertret L, Weissenbach J (1996) Mapping of a second (1996) Congenital recessive ichthyosis unlinked to loci for locus for lamellar ichthyosis to chromosome 2q33–35. Hum epidermal transglutaminases. J Invest Dermatol 107: Molec Genet 5:555–559 808–811 Phillips MA, Stewart BE, Rice RH (1992) Genomic structure Compton JG, DiGiovanna JJ, Santucci SK, Kearns KS, Amos of keratinocyte transglutaminase: recruitment of new exon CI, Abangan DL, Korge BP, et al (1992) Linkage of epider- for modified function. J Biol Chem 267:2282–2286 molytic hyperkeratosis to the type II keratin gene cluster on Russell LJ, DiGiovanna JJ, Hashem N, Compton JG, Bale SJ chromosome 12q. Nat Genet 1:301–305 (1994) Linkage of autosomal recessive lamellar ichthyosis Dib C, Faure´ S, Fizames C, Samson D, Drouot N, Vignal A, to chromosome 14q. Am J Hum Genet 55:1146–1152 Hennies et al.: Genotype/Phenotype Correlation in LI 1061

Russell LJ, DiGiovanna JJ, Rogers GR, Steinert PM, Hashem Traupe H, Kolde G, Happle R (1984) Autosomal dominant N, Compton JG, Bale SJ (1995) Mutations in the gene for lamellar ichthyosis: a new skin disorder. Clin Genet 26: transglutaminase 1 in autosomal recessive lamellar ichthy- 457–461 osis. Nat Genet 9:279–283 Wang M, Kim IG, Steinert PM, McBride OW (1994) Assign- Schuler GD, Boguski MS, Stewart EA, Stein LD, Gyapay G, ment of the human transglutaminase 2 (TGM2) and trans- Rice K, White RE, et al (1996) A gene map of the human glutaminase 3 (TGM3) genes to chromosome 20q11.2. Ge- genome. Science 274:540–546 nomics 23:721–722 Takahashi N, Takahashi Y, Putnam FW (1986) Primary struc- Williams ML, Elias PM (1986) The ichthyoses. In: Thiers BH, ture of blood coagulation factor XIIIa (fibrinoligase, trans- Dobson RL (eds) Pathogenesis of skin disease. Livingston, glutaminase) from human placenta. Proc Natl Acad Sci USA New York, pp 519–555 83:8019–8023 Yamanishi K, Inazawa J, Liew FM, Nonomura K, Ariyama T, Thacher SM, Rice RH (1985) Keratinocyte-specific transglu- Yasuno H, Abe T, et al (1992) Structure of the gene for taminase of cultured human epidermal cells: relation to human transglutaminase 1. J Biol Chem 267:17858–17863 cross-linked envelope formation and terminal differentia- Yee VC, Pedersen LC, Le Trong I, Bishop PD, Stenkamp RE, tion. Cell 40:685–695 Teller DC (1994) Three-dimensional structure of a trans- Traupe H (1989) The ichthyoses: a guide to clinical diagnosis, glutaminase: human blood coagulation factor XIII. Proc genetic counseling, and therapy. Springer, Berlin Natl Acad Sci USA 91:7296–7300